Research On Task Scheduling Strategy Based On Multi-core Processors

With the emergence of multi-core processors, task scheduling for multi-core processors has become one of the focus in high efficient multi-core processors study. For the past few years, many experts and scientific research institute both home and abroad do a lot of research in task scheduling. In order to shorten the scheduling length, they minimize the communication costs that between two tasks and change the order of execution. However, the efficiency of task scheduling presently is very low. The problems such as high communication costs, low priority for crucial task and too much redundant tasks reduce the scheduling efficiency for multi-core processors greatly. In order to enhance the performance of multi-core processors system, it is important to find a new and higher efficiency task scheduling algorithm.Based on the study and analysis of existing task scheduling algorithms for multi-core processors, a high efficient comprehensive task scheduling strategy is proposed in this paper, a new task scheduling algorithm is designed correspondingly. In comparison with these existing task scheduling algorithms, this new task scheduling strategy has obvious advantages as the follows. Firstly, this new algorithm optimizes task graph by using clustering, the communication costs of special tasks is minimized greatly. Secondly, the task priority value is calculated in the case of the whole topological structure of task graph was considered, the crucial task will have higher priority consequently. Then, task insert interval and task duplication are considered in the process of task scheduling, as a result, the efficiency of processor resources will be improved. Lastly, the result of task schedule is optimized effectively, these redundant tasks are reduced and the makespan of all tasks is reduced also. Meanwhile, the optimized tasks scheduling strategy is designed to resolve the problem of parallel task scheduling in multi-DAGs. It is meaningful to enhance the efficiency for multi-core processors and to promote the usage in parallel task scheduling in multi-DAGs for the new algorithm.In order to verify the practicality and efficiency of new task scheduling algorithm, experiment are executed through schedule for task graph instance and performance test respectively. To prove the efficiency of the new algorithm, a certain task graph is scheduled as an example. The performance test verify the practicality and efficiency of new algorithm more justly by setting different performance evaluation parameters and generate a large numbers of DAGs which is different from each other. The results show that compare to old scheduling algorithms, this new algorithm can enhance the task scheduling efficiency obviously, it has excellent application prospect.